ipfKnn: Implements the k-nearest neighbors algorithm

In ipft: Indoor Positioning Fingerprinting Toolset

Description

Implements the k-nearest neighbors algorithm

Usage

ipfKnn(train_fgp, train_pos, k = 3, method = "euclidean",
  weights = "distance", norm = 2, sd = 5, epsilon = 0.001, alpha = 1,
  threshold = 20, FUN = NULL, ...)

Arguments

train_fgp	a data frame containing the fingerprint vectors of the training set
train_pos	a data frame containing the positions of the training set observations
k	the k parameter for knn algorithm (number of nearest neighbors)
method	the method to compute the distance between the RSSI vectors: 'euclidean', 'manhattan', 'norm', 'LGD' or 'PLGD'
weights	the algorithm to compute the weights: 'distance' or 'uniform'
norm	parameter for the 'norm' method
sd	parameter for 'LGD' and 'PLGD' methods
epsilon	parameter for 'LGD' and 'PLGD' methods
alpha	parameter for 'PLGD' method
threshold	parameter for 'PLGD' method
FUN	an alternative function provided to compute the distance. This function must return a matrix of dimensions: nrow(test) x nrow(train), containing the distances from test observations to train observations. The two first parameters taken by the function must be train and test
...	additional parameters for provided function FUN

Value

An S3 object of class ipfModel, with the following properties: params -> a list with the parameters passed to the function data -> a list with the fingerprints and locations

Examples

    model <- ipfKnn(ipftrain[, 1:168], ipftrain[, 169:170], k = 9, method = 'manhattan')

ipft documentation built on May 2, 2019, 7:23 a.m.